Vertex AI Forecasting
This specialized course provides data practitioners with a practical introduction to developing end-to-end forecasting solutions on Google Cloud. Learners work through hands-on labs covering time series data ingestion into managed datasets, building AutoML Forecasting models in Vertex AI, and adding forecasting workflow automation with Vertex Pipelines.
What you will learn:
- Ingest, preprocess, and manage time series data using Vertex AI managed datasets.
- Build and evaluate Vertex AI Forecasting models using AutoML.
- Run batch predictions using the developed forecasting model and generate forecasts.
- Build a complete model retraining pipeline using Vertex Pipelines and pre-built Kubeflow components
Who this course is for?
- This offering is for professional Data Engineers and Data Scientists, and ML Engineers who want to build end-to-end high performance forecasting solutions on Google Cloud. Although there are multiple ways to build forecasting models on Google Cloud, this course specifically focuses on building an AutoML forecasting model using Vertex AI.
Prerequisite
Participants are expected to have:
- Working proficiency with Python on topics covered in the Google Crash Course on Python.
- Working proficiency with data engineering and machine learning on Google Cloud as covered in Google Cloud Big Data and Machine Learning Fundamentals course.
Course Outline
- Module 1: Introduction
- Carry out the steps to launch hands-on labs using Qwiklabs platform.
- Module 2: Introduction Vertex AI Forecasting
- Describe the options supported by Vertex AI to enable the use of machine learning in applications.
- Summarize the important steps in building a Vertex AI Forecasting solution.
- List the various tasks automatically performed by AutoML while building a machine learning model.
- Module 3: Forecasting Foundations
- Contrast statistical approach with the machine learning based approach for building forecasting solutions.
- Enumerate the use cases where the machine learning approach is superior to statistical based approach.
- Select an appropriate workflow between BQ ARIMA+ and Vertex AI Forecasting depending on the various selection factors.
- Module 4: Project Setup
- Describe the various infrastructure components required to build a forecasting solution and to secure it.
- Module 5: Data Ingestion
- Convert the input data into a narrow format data satisfying the requirements specified by Vertex AI Forecasting.
- Create Train, Validate and Test datasets from the input data.
- Explain the best practices that should be followed for the data ingestion stage.
- Module 6: Model Development using AutoML
- Enumerate the input options required to build the Vertex AI Forecasting solution and describe their significance.
- Select an appropriate optimization objective based on the requirements of a use case.
- Module 7: Model Deployment, Predictions and Evaluation
- Judge the performance of the forecasting model by reviewing the model evaluation report.
- Generate model forecasts using batch predictions.
- Module 8: Retraining the Model
- Summarize the advantages of Vertex Pipelines service over the Kubeflow.
- List the reasons that necessitates a frequent retraining of forecasting models.
- Build a complete model retraining pipeline using Vertex Pipelines and pre-built Kubeflow components.
- Module 9: Course Conclusion
- Summarize important steps in an AutoML forecasting solution using Vertex AI.
Jadwal Training
| Tanggal | Pukul | Biaya (per pax; belum termasuk VAT 10%) | Trainer | Venue | Daftar |
|---|---|---|---|---|---|
| TBA | TBA | Rp 19 juta | Satria Yuda Utama | Online | TBA |
| TBA | TBA | Rp 19 juta | Satria Yuda Utama | Online | TBA |
| TBA | TBA | Rp 19 juta | Satria Yuda Utama | Online | TBA |
| TBA | TBA | Rp 19 juta | Satria Yuda Utama | Online | TBA |
| TBA | TBA | Rp 19 juta | Satria Yuda Utama | Online | TBA |